This project will develop and validate a tool for increasing manual precision for retinal microsurgery. The proposed project will combine "Micron," a handheld tremor-reducing instrument (developed at CMU) with stereo image-processing and tracking software (developed at JHU) to create an active handheld image guided interventional device for microsurgery. This device will help to guide the surgeon to a target using the tip-deflection capability of Micron. This device will be tested in vivo on chick chorioallantoic membrane to determine its efficacy in retinal vein cannulation. The three specific aims of this proposal are: 1. Development of vision-based instrument tracking algorithms: Precise intervention requires knowledge of the location of the target on the surface of the retina, and knowledge of the location of the surgical tool relative to that target. This aim therefore involves the development of algorithms to locate and visually track the target and tool in stereo microscope images. 2. Development of instrument tip stabilization algorithms: Given the relative location of the tool tip and target in a stereo image pair, control algorithms will be developed to manipulate the tip of Micron toward the target, while still allowing the surgeon holding the instrument to retain final control of its tip location. This aim therefore involves the development of algorithms for controlling the tool tip from visual information. 3. Evaluation of the effectiveness of active tip guidance: The enhanced Micron device will be tested by trained surgeons to determine if image-based guidance improves microsurgical performance. Retinal vein cannulation will be used as the benchmark task. Retinal vein occlusion occurs at the rate of roughly 100,000 new cases per year, and there is currently no widely used, effective treatment. The proposed device will facilitate the development of treatments for this disease as well as other microsurgical procedures that are similarly infeasible using currently available tools and techniques. [unreadable] [unreadable] [unreadable]